fix: make delete orphan job compatible with sharding

Signed-off-by: Robin Appelman <robin@icewind.nl>

apps/files/lib/Command/DeleteOrphanedFiles.php

@@ -36,33 +36,27 @@ protected function configure(): void {
 	public function execute(InputInterface $input, OutputInterface $output): int {
 		$deletedEntries = 0;
 
-		$query = $this->connection->getQueryBuilder();
-		$query->select('fc.fileid')
-			->from('filecache', 'fc')
-			->where($query->expr()->isNull('s.numeric_id'))
-			->leftJoin('fc', 'storages', 's', $query->expr()->eq('fc.storage', 's.numeric_id'))
-			->setMaxResults(self::CHUNK_SIZE);
+		$deletedStorages = array_diff($this->getReferencedStorages(), $this->getExistingStorages());
+
+		$deleteExtended = !$input->getOption('skip-filecache-extended');
+		if ($deleteExtended) {
+			$fileIdsByStorage = $this->getFileIdsForStorages($deletedStorages);
+		}
 
 		$deleteQuery = $this->connection->getQueryBuilder();
 		$deleteQuery->delete('filecache')
-			->where($deleteQuery->expr()->eq('fileid', $deleteQuery->createParameter('objectid')));
+			->where($deleteQuery->expr()->in('storage', $deleteQuery->createParameter('storage_ids')));
 
-		$deletedInLastChunk = self::CHUNK_SIZE;
-		while ($deletedInLastChunk === self::CHUNK_SIZE) {
-			$deletedInLastChunk = 0;
-			$result = $query->execute();
-			while ($row = $result->fetch()) {
-				$deletedInLastChunk++;
-				$deletedEntries += $deleteQuery->setParameter('objectid', (int) $row['fileid'])
-					->execute();
-			}
-			$result->closeCursor();
+		$deletedStorageChunks = array_chunk($deletedStorages, self::CHUNK_SIZE);
+		foreach ($deletedStorageChunks as $deletedStorageChunk) {
+			$deleteQuery->setParameter('storage_ids', $deletedStorageChunk, IQueryBuilder::PARAM_INT_ARRAY);
+			$deletedEntries += $deleteQuery->executeStatement();
 		}
 
 		$output->writeln("$deletedEntries orphaned file cache entries deleted");
 
-		if (!$input->getOption('skip-filecache-extended')) {
-			$deletedFileCacheExtended = $this->cleanupOrphanedFileCacheExtended();
+		if ($deleteExtended) {
+			$deletedFileCacheExtended = $this->cleanupOrphanedFileCacheExtended($fileIdsByStorage);
 			$output->writeln("$deletedFileCacheExtended orphaned file cache extended entries deleted");
 		}
 
@@ -72,28 +66,63 @@ public function execute(InputInterface $input, OutputInterface $output): int {
 		return self::SUCCESS;
 	}
 
-	private function cleanupOrphanedFileCacheExtended(): int {
-		$deletedEntries = 0;
-
+	private function getReferencedStorages(): array {
 		$query = $this->connection->getQueryBuilder();
-		$query->select('fce.fileid')
-			->from('filecache_extended', 'fce')
-			->leftJoin('fce', 'filecache', 'fc', $query->expr()->eq('fce.fileid', 'fc.fileid'))
-			->where($query->expr()->isNull('fc.fileid'))
-			->setMaxResults(self::CHUNK_SIZE);
+		$query->select('storage')
+			->from('filecache')
+			->groupBy('storage')
+			->runAcrossAllShards();
+		return $query->executeQuery()->fetchAll(\PDO::FETCH_COLUMN);
+	}
 
-		$deleteQuery = $this->connection->getQueryBuilder();
-		$deleteQuery->delete('filecache_extended')
-			->where($deleteQuery->expr()->in('fileid', $deleteQuery->createParameter('idsToDelete')));
+	private function getExistingStorages(): array {
+		$query = $this->connection->getQueryBuilder();
+		$query->select('numeric_id')
+			->from('storages')
+			->groupBy('numeric_id');
+		return $query->executeQuery()->fetchAll(\PDO::FETCH_COLUMN);
+	}
 
-		$result = $query->executeQuery();
-		while ($result->rowCount() > 0) {
-			$idsToDelete = $result->fetchAll(\PDO::FETCH_COLUMN);
+	/**
+	 * @param int[] $storageIds
+	 * @return array<int, int[]>
+	 */
+	private function getFileIdsForStorages(array $storageIds): array {
+		$query = $this->connection->getQueryBuilder();
+		$query->select('storage', 'fileid')
+			->from('filecache')
+			->where($query->expr()->in('storage', $query->createParameter('storage_ids')));
+
+		$result = [];
+		$storageIdChunks = array_chunk($storageIds, self::CHUNK_SIZE);
+		foreach ($storageIdChunks as $storageIdChunk) {
+			$query->setParameter('storage_ids', $storageIdChunk, IQueryBuilder::PARAM_INT_ARRAY);
+			$chunk = $query->executeQuery()->fetchAll();
+			foreach ($chunk as $row) {
+				$result[$row['storage']][] = $row['fileid'];
+			}
+		}
+		return $result;
+	}
 
-			$deleteQuery->setParameter('idsToDelete', $idsToDelete, IQueryBuilder::PARAM_INT_ARRAY);
-			$deletedEntries += $deleteQuery->executeStatement();
+	/**
+	 * @param array<int, int[]> $fileIdsByStorage
+	 * @return int
+	 */
+	private function cleanupOrphanedFileCacheExtended(array $fileIdsByStorage): int {
+		$deletedEntries = 0;
 
-			$result = $query->executeQuery();
+		$deleteQuery = $this->connection->getQueryBuilder();
+		$deleteQuery->delete('filecache_extended')
+			->where($deleteQuery->expr()->in('fileid', $deleteQuery->createParameter('file_ids')));
+
+		foreach ($fileIdsByStorage as $storageId => $fileIds) {
+			$deleteQuery->hintShardKey('storage', $storageId, true);
+			$fileChunks = array_chunk($fileIds, self::CHUNK_SIZE);
+			foreach ($fileChunks as $fileChunk) {
+				$deleteQuery->setParameter('file_ids', $fileChunk, IQueryBuilder::PARAM_INT_ARRAY);
+				$deletedEntries += $deleteQuery->executeStatement();
+			}
 		}
 
 		return $deletedEntries;

apps/files/tests/Command/DeleteOrphanedFilesTest.php

@@ -64,13 +64,19 @@ protected function tearDown(): void {
 	}
 
 	protected function getFile($fileId) {
-		$stmt = $this->connection->executeQuery('SELECT * FROM `*PREFIX*filecache` WHERE `fileid` = ?', [$fileId]);
-		return $stmt->fetchAll();
+		$query = $this->connection->getQueryBuilder();
+		$query->select('*')
+			->from('filecache')
+			->where($query->expr()->eq('fileid', $query->createNamedParameter($fileId)));
+		return $query->executeQuery()->fetchAll();
 	}
 
 	protected function getMounts($storageId) {
-		$stmt = $this->connection->executeQuery('SELECT * FROM `*PREFIX*mounts` WHERE `storage_id` = ?', [$storageId]);
-		return $stmt->fetchAll();
+		$query = $this->connection->getQueryBuilder();
+		$query->select('*')
+			->from('mounts')
+			->where($query->expr()->eq('storage_id', $query->createNamedParameter($storageId)));
+		return $query->executeQuery()->fetchAll();
 	}
 
 	/**

lib/private/DB/QueryBuilder/ExtendedQueryBuilder.php

@@ -289,8 +289,8 @@ public function executeStatement(?IDBConnection $connection = null): int {
 		return $this->builder->executeStatement($connection);
 	}
 
-	public function hintShardKey(string $column, mixed $value) {
-		$this->builder->hintShardKey($column, $value);
+	public function hintShardKey(string $column, mixed $value, bool $overwrite = false) {
+		$this->builder->hintShardKey($column, $value, $overwrite);
 		return $this;
 	}
 

lib/private/DB/QueryBuilder/QueryBuilder.php

@@ -1366,7 +1366,7 @@ public function quoteAlias($alias) {
 		return $this->helper->quoteColumnName($alias);
 	}
 
-	public function hintShardKey(string $column, mixed $value) {
+	public function hintShardKey(string $column, mixed $value, bool $overwrite = false) {
 		return $this;
 	}
 

lib/private/DB/QueryBuilder/Sharded/ShardedQueryBuilder.php

@@ -285,7 +285,11 @@ private function registerOrder(string $column, string $order): void {
 		];
 	}
 
-	public function hintShardKey(string $column, mixed $value) {
+	public function hintShardKey(string $column, mixed $value, bool $overwrite = false) {
+		if ($overwrite) {
+			$this->primaryKeys = [];
+			$this->shardKeys = [];
+		}
 		if ($this->shardDefinition?->isKey($column)) {
 			$this->primaryKeys[] = $value;
 		}

lib/public/DB/QueryBuilder/IQueryBuilder.php

@@ -1027,7 +1027,7 @@ public function getColumnName($column, $tableAlias = '');
 	 * @return $this
 	 * @since 30.0.0
 	 */
-	public function hintShardKey(string $column, mixed $value);
+	public function hintShardKey(string $column, mixed $value, bool $overwrite = false);
 
 	/**
 	 * Set the query to run across all shards if sharding is enabled.